§1.1 Field of the Invention
The present invention concerns avoiding data traffic loss in an Ethernet Ring multihomed, in an active-standby manner, to a virtual private LAN service (VPLS) transport network. Thus, the present invention may be used, for example, to prevent data losses in Border Gateway Protocol (BGP) multihomed Ethernet Rings, multi-chassis-lag (MC-LAG) multihomed Ethernet Rings, and other types of active-standby multihomed Ethernet Rings.
§1.2 Background Information
§1.2.1 Network Communications Protection Using Rings
Link failure may often be an unavoidable part of networking. However, there are techniques for improving the reliability of a router or bridge network even when link failures occur. For example, SONET/SDH seal-healing rings may be used to add a level of robustness to communications networks. Such ring protection switching has been extended to Ethernet links. Ethernet Ring Protection (ERP) may be configured for a series of two or more systems so that if one link fails, traffic is rerouted around the failure on the Ethernet Ring (ring). ERP switching architectures avoid loops, and use learning, forwarding, and Filtering Database (FDB) mechanisms. Each of these aspects of ERP switching architectures are introduced below.
Loop avoidance in the ring may be achieved by permitting data traffic flow on all but one of the links in the ring, at any point in time. The particular link avoided may be referred to as the Ring Protection Link (RPL). Under normal conditions this RPL may be blocked or otherwise disabled (i.e., not used for data traffic). A designated ring node, referred to as the RPL Owner Node, may be responsible for blocking traffic through the RPL under normal conditions. Under a ring failure condition, the RPL Owner Node is responsible for unblocking (i.e., activating) the RPL (unless the RPL has failed), allowing the RPL to be used for data traffic.
Thus, ERP uses one specific link (for example, the RPL) to protect the whole ring. As illustrated in FIG. 1A, the example Ethernet Ring 100 includes an RPL 105 to protect it. When all links in the ring are up and active, data traffic through the RPL 105 is blocked. Therefore, the RPL 105 remains idle. Referring to FIGS. 1A and 1B, although the physical topology 110 of the Ethernet Ring 100 includes a physical RPL 105, in the logical topology 110′ of Ethernet Ring 100, the RPL 104 is not seen between “Node 2” and the “RPL Owner” node 115. FIG. 1B illustrates the condition of the RPL being idle, under which no data traffic is allowed to pass through the RPL. In this example, the RPL 105 is controlled by a designated RPL owner node 115. The RPL owner node 115 may be responsible for blocking the RPL under normal operating conditions.
Referring to FIG. 1C, if a link failure occurs on the ring (e.g., if the link between Nodes 4 and 5 fails), the RPL owner node 115 automatically protects the ring by unblocking the RPL 105 so that the data traffic can be forwarded on an alternate path around the ring. For example, as shown in logical topology 120 of Ethernet Ring 100 in FIG. 1C, when the link between “Node 4” and “Node 5” fails, the RPL 105 is activated by the “RPL Owner” node 115 and data traffic is allowed to pass through the RPL 105. If the link between “Node 4” and “Node 5” recovers from failure, the RPL owner node 115 may, responsive to this recovery, revert the ring to the normal condition (Recall logical topology 110′ of FIG. 1B.) by deactivating the RPL 105 and blocking the data traffic through the RPL 105.
Enhanced ring protection (ERP) works on the basis of a filtering data base (FDB) flush. More specifically, upon protection switching for a failure or a failure recovery (that is, when the ring switches from normal condition to failure condition, or vice-versa) all nodes of the ring remove all learned Media Access Control (MAC) addresses in their FDBs for a changed ring topology. Then, each ring node of the ring may broadcast data frames until MAC address learning of nodes of a newly configured ring is completed.
§1.2.2 Network Communications Protection Using Multihoming
“Multihoming” is a technique used to increase the reliability of an Internet connection for an IP network. There are various techniques that provide active-standby multihoming of an Ethernet Ring to a VPLS transport network. As one example, Border Gateway Protocol (BGP) multihoming enables a customer site to be connected with a service provider network via two or more peer Provider Edge (PE) routers (for example, border routers running BGP). Multi-Chassis-Lag (MC-LAG) is another example of a VPLS multihoming technique that provides Active-Standby multihoming. In either case (or in some other type active-standby multihoming of an Ethernet Ring to a transport network), the service provider may be a network that provides Virtual Private LAN Service (VPLS), for example. Connecting the customer site to two or more Provider Edge (PE) routers provides redundant connectivity that maintains the VPLS and traffic forwarding to and from the multihomed site in the event of PE router-to-Consumer Edge (CE) device link failure, the failure of a PE router, the failure of a CE device, or a Multi Protocol Label Switching (MPLS) reachability failure between a local PE router and a remote PE router. A redundant (backup) CE device-to-PE edge router path may begin providing service to the customer site responsive to the detection of one of the foregoing failures.
§1.2.3 Challenges Protecting Network Communications when an Ethernet Protection Ring Uses Multihoming
Referring to FIG. 2, consider an architecture 200 in which a customer site 205 accesses PE routers 230 and 235 via CE routers CE 1 and CE 2 that are arranged in an Ethernet Protection Ring topology 265. More specifically, in the example architecture 200, the customer site 205 may be multihomed by connecting one of the CE routers (e.g., CE 1, referred to as a first node) in the Ethernet Ring to a primary PE router (e.g., PE 1 230) and another one of the CE routers (e.g., CE 2, referred to as a second node) to a standby PE router (e.g., PE 2 235). Under such an arrangement, a failure of the link between the primary PE router and the first node CE router (CE 1-PE 1) will not cause the Ethernet Ring to switch to a mode in which a protection link is activated because there is no failure sensed in the Ethernet Ring. As an unfortunate consequence, the Ethernet Ring likely will not forward the data traffic to the second CE router, which is connected to the standby PE router (PE 2 235). This may cause “black holing” of the data packets sent to the primary PE router. That is, the data packets may be lost or dropped without the sender knowing.
In view of the foregoing, it would be useful to extend ERP to an Ethernet Ring that is multihomed, in an active-standby manner, to a VPLS transport network (such as to a BGP multihomed Ethernet Ring, an MC-LAG multihomed Ethernet Ring, or some other type of active-standby multihomed Ethernet Ring, etc.) that protects both the Ethernet Ring and the connection to the service provider (e.g., providing a VPLS), thereby avoiding data traffic loss in the multihomed Ethernet Ring.